Dear Max,
thanks for your valuable comment. I assume that you used the function
for regression - not classification.
I use Mac OS X plattform (version 10.5.6). The R version is 2.8.1 (I
prefer to update to 2.9.1 not 2.9.0). The kernlab package version is
0.9-8.
The x and y-input into LSSVM regression can be reproduced by:
x<-<-
matrix
(data
=
rnorm
(n
=
12800
,mean
=0.0053,sd=0.0146),nrow=100,ncol=128,dimnames=list(c(1:100),c(1:128)));
y<-rnorm(n=100,mean=0.7958,sd=0.1900);
The function is:
test
<
_lssvm
(test
,test.ref,scale=F,type="regression",kernel="rbfdot",kpar=list(sigma=5));
The error message is:
Error in if (n != dim(y)[1]) stop("Labels y and data x dont match") :
argument is of length zero.
This error message does not make sense to me. The R documentation for
the lssvm function states that y can be a vector. The dimension of a
vector is always NULL. Thus, the check function in the error message
[if(n !=dim(y)[1]) stop] does not add up for me.
Regards,
Thomas
On 14 May 2009, at 15:33, Max Kuhn wrote:
>> To make things easier (using only two optimization parameters and not
>> loosing performance) I wanted to use LS SVM regression
>> (lssvm{kernlab}). But
>> it looks to me that it is not yet implemented. At least I got error
>> messages, which I could not find a solution for (Error in if (n !
>> _dim(y)[1]
>> stop ("Labels y and data x dont match").
>
> I've used the lssvm function in kernlab without issue.
>
> You should follow the posting guide and provide a reproducible example
> so that there is a possibility of answering your question. Plus, what
> versions etc.
>
> Max
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
